---
title: Semana 
date: "`r Sys.Date()`"

output: 
  flexdashboard::flex_dashboard:
    theme: sandstone
    social: menu
    source_code: embed
    #orientation: rows
    #orientation: columns
    vertical_layout: fill
---


```{r setup, include=FALSE}
library(flexdashboard)
library(yaml)
library(knitr)
require(plotly)
library(ggplot2)
require(gganimate)
require(esquisse)
require(gapminder)
library(ggplot2)
library(kableExtra)
library(animation)

##### LIBRERIAS
library(sf)
library(sp)
library(raster)
library(leaflet)
library(tidyverse)
library(ggmap)
library(reshape2)
library(dplyr)
library(spatstat)
library(tmaptools)
library(readr)
#library(gifski)
#library(tibble)
library(lubridate)
library(hms)
library(leaflet.extras)
#library(magrittr)

library(foreign)
library(xtable)
library(base)
library(DT)
#library(data.table)
#library(forecast)
# library(rgdal)
# library(spbabel)
library(maptools)
library(tmap)
library(DT)
library(reshape2)
library(reshape)
```


```{r}

#dat = read.table('datos_app_15dias/datos_app_15dias.csv',sep = ',',header = T)

dat = read.table('waze_15dias.csv',sep = ',',header = T)
#datosapp15

dat = dat %>% filter(city == "Montevideo")


datJam = dat %>% mutate(datecreated = as.POSIXct(strptime(gsub('T',' ',datecreated), "%Y-%m-%d %H:%M:%S")),
                               datemodified = as.POSIXct(strptime(gsub('T',' ',datemodified), "%Y-%m-%d %H:%M:%S")),
                        time_index = as.POSIXct(strptime(gsub('T',' ',time_index), "%Y-%m-%d %H:%M:%S")),
                               datepublished = as.POSIXct(strptime(gsub('T',' ',datepublished), "%Y-%m-%d %H:%M:%S")))


datJam = datJam %>% filter(datemodified>min(datJam$datemodified)+60*60*24)
```


```{r}

df <- datJam %>% dplyr::select(datemodified,street,length,delay,speed,speedkmh,level,location_centroid_lat,location_centroid_lon,datemodified,datecreated)


datJam <- df
```


```{r}
#extrae coordenadas del centroide
datJam$Xc <- datJam$location_centroid_lat
datJam$Yc <- datJam$location_centroid_lon
```


```{r}
#hago las transformaciones para dia hora mes, etc

datJam = datJam %>% 
  mutate(dia = day(datJam$datemodified),
         mes = month(datJam$datemodified),
         ano = year(datJam$datemodified),
         hora = hour(datJam$datemodified),
         NomDia = wday(datJam$datemodified),
         MesDia = ifelse(nchar(mes)==2, paste0(mes,'-',dia),paste0('0',mes,'-',dia)),
         #DiaMes = ifelse(nchar(mes)==2, paste0(mes,'-',dia),paste0(mes,'-0',dia)),
        #DiaMesHora = paste0(DiaMes,'-',hora))
        MesDiaHora = ifelse(nchar(hora)==2,paste0(MesDia,'-',hora),paste0(MesDia,'-0',hora)))

# MesDia = ifelse(nchar(datJam$mes)==2, paste0(datJam$mes,'-',datJam$dia),paste0('0',datJam$mes,'-',datJam$dia))

datJam$nombreDia <- ifelse(datJam$NomDia == 1, 'Domingo',
                   ifelse(datJam$NomDia == 2, 'Lunes',
                          ifelse(datJam$NomDia == 3, 'Martes',
                                 ifelse(datJam$NomDia == 4, 'Miercoles',
                                        ifelse(datJam$NomDia == 5, 'Jueves',
                                               ifelse(datJam$NomDia == 6, 'Viernes',
                                                      ifelse(datJam$NomDia == 7, 'Sabado',0)))))))


datJam$nombreDia <- factor(datJam$nombreDia
            ,ordered = TRUE,
            levels = c("Domingo", "Lunes", "Martes", "Miercoles",'Jueves','Viernes','Sabado'))


datJam = datJam %>% 
  mutate(NomMesDia = paste0(MesDia ,'-',nombreDia))

#min(datJam$datemodified)

```



```{r}

#trabajo con semanas
undia <- 1*24*60*60
unasemana <- undia*7

sem1 <- datJam %>% filter(datemodified< (min(datJam$datemodified)+unasemana))
sem2 <-  datJam %>% filter(datemodified>= (min(datJam$datemodified)+unasemana))

# sem1 <- datJam %>% filter(datemodified > (min(datJam$datemodified) + undia) & (datemodified < min(datJam$datemodified)+unasemana+undia))
# 
# sem2 <-  datJam %>% filter((datemodified> min(datJam$datemodified)+unasemana+undia) & (datemodified < max(datJam$datemodified)))


datJam$sem <- ifelse(datJam$datemodified < (min(datJam$datemodified)+unasemana),'sem1','sem2')

#datJam <- datJam %>% filter(datemodified > min(datJam$datemodified)+undia)

```



**Descripción** 
=======================================================================

Column {data-width=400}
-----------------------------------------------------------------------




### **Descripción**

Waze genera información sobre atascos de tráfico procesando las siguientes fuentes de datos:

- Ubicación de GPS enviados de los teléfonos de los usuarios que conducen mientras usan la aplicación, información generada en tiempo real en funciónde la ubicación y velocidad. 
- Cálculo de velocidad real frente a la velocidad media (en un intervalo de tiempo específico) y flujo libre (velocidad máxima media en el segmento)

- Reportes compartidos por los usuarios de Waze que están en un atasco   

Descripción de algunas variables utilizadas en este análisis:

- **delay**: Retraso con respecto a un flujo libre en segundos
- **level**: Nivel de atasco (0 a 5, 0 libre y 5 muy congestionado)
- **length**: Largo en metros
- **speedkmh**: Velocidad promedio en kilómetros por hora

**Semana actual: del `r min(sem2$datemodified)` al día `r max(sem2$datemodified)`**  
**Semana anterior: del `r min(sem1$datemodified)` al día `r max(sem1$datemodified)`** 


Column {data-width=400}
-----------------------------------------------------------------------


### **Cantidad de eventos por día**




```{r}
resumen <- datJam %>%
  filter(sem == 'sem2') %>%
  group_by(level) %>%
  dplyr::summarise(Cant=n())

Total <- sum(resumen$Cant)

resumen2 <- rbind(resumen, c("Total",Total))


# 
# resumen1 <- datJam %>%
#   filter(sem == 'sem1') %>%
#   group_by(level) %>%
#   dplyr::summarise(Cant=n())
# 
# Total1 <- sum(resumen1$Cant)
# 
# resumen1 <- rbind(resumen1, c("Total1",Total1))
# 
# resumenT <- datJam %>%
#   #filter(sem == 'sem1') %>%
#   group_by(level) %>%
#   dplyr::summarise(Cant=n())
# 
# TotalT <- sum(resumenT$Cant)
# 
# resumenT <- rbind(resumenT, c("TotalT",TotalT))
# kable(resumen1) %>% 
#   kable_styling(bootstrap_options = c("striped", "hover", "condensed"))
  
```

```{r}
x = sem2 %>%
  group_by(NomMesDia) %>%
  dplyr::summarise(n_dM=n()) %>%
  mutate(prop = round(100*prop.table(n_dM),1)) %>%
  ungroup()

#x <- x[x$NomMesDia != '06-2-Jueves',]

evol_md_2 <- ggplot(x) +
  aes(x=NomMesDia, y=n_dM) +
  geom_bar(stat = "identity", fill="darkgrey") +
  theme_minimal()+
  theme(legend.position = 'top',
                            legend.title = element_blank(),
                            axis.text.x = element_text(angle = 45,hjust = 1,vjust = 1,size = 7)) + 
  # geom_text(aes(label=n_dM), vjust=0, color="black",    #15
  #             position = position_dodge(0.9),  size=2.0
  #           ) +
  # geom_text(aes(label=prop), vjust = 1, color="black",    #15
  #              position = position_dodge(0.9),  size=2.0
  #            ) +
  labs(x = "Fecha", y = 'Cantidad de eventos', colour = "Cantidad eventos")

# title = "Congestion diaria por dia de la semana actual (todos los niveles)", 
ggplotly(evol_md_2)



```



### **Cantidad de eventos por nivel**

```{r}
x_level = sem2 %>%
  group_by(level) %>%
  dplyr::summarise(n_dM=n()) %>%
  mutate(prop = round(100*prop.table(n_dM),1)) %>%
  ungroup()

gg_level <- ggplot(x_level) +
  aes(x=level, y=n_dM) +
  geom_bar(stat = "identity", fill="darkgrey") +
  theme_minimal()+
  theme(legend.position = 'top',
                            legend.title = element_blank(),
                            axis.text.x = element_text(hjust = 1,vjust = 1,size = 7)) + 

  labs(x = "Level", y = 'Cantidad de eventos')

ggplotly(gg_level)

```






























Column {data-width=500}
-----------------------------------------------------------------------

### **Mapa de calor semana actual**

```{r message=FALSE, warning=FALSE, include=FALSE}

#levanto capa municipios
municipios <- st_read("SHP/sig_municipios/sig_municipios.shp")

#levanto capa de barrios
barrios <- st_read("SHP/barrios.gpkg")


mun = read_sf('SHP/sig_municipios/sig_municipios.shp')

ccz = read_sf('SHP/sig_comunales/sig_comunales.shp')
st_crs(ccz) = st_crs(mun)

#paso a coordenadas lat long
st_geometry(mun) = st_transform(st_geometry(mun), 4326)
st_geometry(ccz) = st_transform(st_geometry(ccz), 4326)
```


```{r message=FALSE, warning=FALSE, include=FALSE}
####### ETIQUETAS MUNICIPIOS
etiqMuni <- paste0("Nombre municipio: ", 
                mun$MUNICIPIO)

############ EJEMPLO 1 (BASICO)
#defino una variable numérica para cada municipio (del 1 al 8) pa después usarla pa colorearlos
mun$MUNI <- ifelse(mun$MUNICIPIO == 'A', 1, 
                   ifelse(mun$MUNICIPIO == 'B', 2,
                          ifelse(mun$MUNICIPIO == 'C', 3,
                                 ifelse(mun$MUNICIPIO == 'CH', 4, 
                                        ifelse(mun$MUNICIPIO == 'D', 5, 
                                               ifelse(mun$MUNICIPIO == 'E', 6,
                                                      ifelse(mun$MUNICIPIO == 'F', 7, ifelse(mun$MUNICIPIO == 'G', 8, 0))))))))


###### DEFINO colores
pal <- colorNumeric(
  palette = "YlGn",
  domain = mun$MUNI)


```


```{r message=FALSE, warning=FALSE, include=FALSE}
#mapa con todos los niveles semana 1

mapa_heat_sem2 <- leaflet() %>% # ABRE LA VENTANA PARA HACER EL MAPA
  addTiles(group = "OSM") %>% # DEFINE UN FONDO (POR DEFECTO OSM)
  addProviderTiles(providers$CartoDB.Positron, group = 'CartoDB.Positron') %>%
  # PONE OTRO FONDO PARA ELEGIR
  addPolygons(data=mun, #fillColor = ~pal(mun$MUNI), # AGReGA LA CAPA DE POLIGONAS DE LOS MUNICIPIOS
              
              fillOpacity = 0, 
              weight = 1, #smoothFactor = 0.2,
               popup=etiqMuni,group = 'Municipio') %>%
  addHeatmap(lng=sem2$Xc,lat=sem2$Yc,max=100,radius=5,blur=10) %>%#
  
  #addCircles(niv34$Xc, niv34$Yc,radius = 1,group = 'dia',color = "dia") %>% ##AGREGO LAS ESCUELAS PUBLICAS
    addLayersControl( 
    baseGroups = c("CartoDB.Positron","OSM"),
    overlayGroups = c("congest","Municipio"), # ESTAS SE PUEDEN PRENDER DE A UNA
    options = layersControlOptions(collapsed = FALSE))



#leaflet::addLegend( position = "topright", pal = pal, values = ~mag, group = "Points", title = "Магитуда землетрясений" )
```
  
```{r}
mapa_heat_sem2
```
  






















































































































































































**Comparación diaria**
=======================================================================

Column {data-width=700}
-----------------------------------------------------------------------




### **Congestión diaria por hora semana actual vs semana anterior**

```{r}
semt<-datJam %>%
  group_by(sem,nombreDia,hora) %>%
  dplyr::summarise(Cant=n())


aa_hora_linea1 <- ggplot(semt) +
  aes(x = hora, y = Cant, fill = sem) +
  geom_bar(stat = 'identity',position = 'dodge',alpha=0.5) + 
  theme(legend.position = "top") +
  
  #geom_line() +
    theme_minimal()+
  scale_fill_manual(values=c("dark green","dark orange")) +
  facet_grid(nombreDia~.)+ 
  
  labs(x = "dia_hora", y = 'Cantidad de eventos')

ggplotly(aa_hora_linea1)

```

Column {data-width=300}
-----------------------------------------------------------------------




### **Congestión diaria: variación**


```{r}
#tabla eventos por día y semana

sem_dia<-datJam %>%
  group_by(sem,nombreDia) %>%
  dplyr::summarise(Cant=n())

aa_hora_linea1 <- ggplot(sem_dia, aes(x = nombreDia, y = Cant, fill = sem)) +
  geom_bar(stat = 'identity',position = 'dodge') + 
  #geom_line() +
    theme_minimal()+
  #facet_grid(sem~.)+ 
  labs(x = "dia_hora", y = 'Cantidad de eventos', colour = "Cantidad eventos")

#ggplotly(aa_hora_linea1)

cast_sem_dia = cast(sem_dia, nombreDia~sem)

Total <- cast_sem_dia %>%
  colSums()

cast_sem_diaT = bind_rows(cast_sem_dia, Total)

cast_sem_diaT$nombreDia <- as.character(cast_sem_diaT$nombreDia)
cast_sem_diaT[(dim(cast_sem_diaT)[1]), 1] = "Total"


#cast_sem_dia$total = cast_sem_dia$sem2+ cast_sem_dia$sem1
cast_sem_diaT$variacion = round(100*(cast_sem_diaT$sem2-cast_sem_diaT$sem1)/cast_sem_diaT$sem1,1)

#rownames(cast_sem_dia) <- NULL
colnames(cast_sem_diaT) <-c("Día","Sem_anterior","Sem_actual","Variación")


kable(cast_sem_diaT) %>% 
  kable_styling(bootstrap_options = c("striped", "hover", "condensed"))



```





























**Comparación: todos los niveles**
=======================================================================

Column {data-width=300}
-----------------------------------------------------------------------


### **Cantidad de eventos y variación con respecto a las calles más congestionadas la semana anterior**

```{r}


df_aux = datJam %>% 
  group_by(street,sem) %>%
  dplyr::summarise(Cant=n()) 

#df_aux1 <- df_aux[order(df_aux$Cant, decreasing = TRUE), ]
aa = df_aux %>% 
  group_by(street) %>% 
  slice_max(Cant) %>% 
  arrange(desc(Cant)) %>% 
  head(40)

bb = df_aux %>% inner_join(aa %>% dplyr::select(street))

cast_data = cast(bb, street~sem, sum)
#cast_data = cast(df_aux, street~sem, sum)

# cast_data$variacion = round(100*(cast_data$sem2-cast_data$sem1)/cast_data$sem1,1)
# 
#lo ordeno en funcion de la cantidad de eventos de la semana anterior 
varia_calles <- cast_data[order(cast_data$sem1, decreasing = TRUE), ]
# 
rownames(varia_calles) <- NULL
# 
# kable(head(varia_calles,20)) %>% 
#   kable_styling(bootstrap_options = c("striped", "hover", "condensed"))

tabla <- cast_data %>%
  mutate(variacion = round(100*(sem2-sem1)/sem1,1)) %>%
  arrange(desc(sem1)) %>%
  head(100) %>%
  DT::datatable()

tabla



```

























Col
-----------------------------------------------------------------------


### **Comparación calles más congestionadas: todos los niveles **


```{r}
Cant_calles <- 5

calles<-varia_calles[1:Cant_calles,1]

rank1 <- ggplot() +               
  geom_bar(data = varia_calles, aes(x=reorder(street,sem1), y=sem1),stat = "identity", position = position_dodge(),fill = "dark green",alpha=0.5)+
  geom_bar(data = varia_calles, aes(x=street, y=sem2),stat = "identity", position = position_dodge(),alpha=0.4)+
    coord_flip() +
  theme_minimal()+
  labs(x= NULL, y = "Cantidad de eventos")  #+
  #ggtitle("Ranking de calles más congestionadas según semana anterior")

ggplotly(rank1)

#gg <- ggplotly(rank1, tooltip = c("Street", "Cant")) 

#highlight(gg, on = "plotly_click")
         
##ggplotly por defecto pone en la etiqueta lo que está en aes         
#tooltip = c("Calle", "Cantidad") selecciona parte de las etiquetas
```












```{r}


aa_hora_linea <- ggplot(semt, aes(x = hora, y = Cant, color = nombreDia)) +
  geom_line() +
  theme_minimal()+
  facet_grid(sem~.)+
  labs(title = "Congestion diaria por hora semana actual vs semana anterior (OTRA VERS)", x = "dia_hora", y = 'Cantidad de eventos', colour = "Cantidad eventos")
#ggplotly(aa_hora_linea)

```








**Comparación: Niveles 3 y 4**
=======================================================================

Column {data-width=300}
-----------------------------------------------------------------------

### **Cantidad de eventos y variación con respecto a las calles más congestionadas la semana anterior**


```{r}
# library(reshape2)
# library(reshape)


df_aux34 = datJam %>% 
  filter(datJam$level>2) %>%
  group_by(street,sem) %>%
  dplyr::summarise(Cant=n()) 

#df_aux1 <- df_aux[order(df_aux$Cant, decreasing = TRUE), ]
aa34 = df_aux34 %>% 
  group_by(street) %>% 
  slice_max(Cant) %>% 
  arrange(desc(Cant)) %>%
  head(40)

bb34 = df_aux34 %>% inner_join(aa34 %>% dplyr::select(street))

cast_data34 = cast(bb34, street~sem, sum)
#cast_data = cast(df_aux, street~sem, sum)

# cast_data34$variacion = round(100*(cast_data34$sem2-cast_data34$sem1)/cast_data34$sem1,1)
# 
# #lo ordeno en funcion de la cantidad de eventos de la semana anterior 
varia_calles34 <- cast_data34[order(cast_data34$sem1, decreasing = TRUE), ]
# 
rownames(varia_calles34) <- NULL

# kable(head(varia_calles34,20)) %>% 
#   kable_styling(bootstrap_options = c("striped", "hover", "condensed"))


tabla34 <- cast_data34 %>%
  mutate(variacion = round(100*(sem2-sem1)/sem1,1)) %>%
  arrange(desc(sem1)) %>%
  head(100) %>%
  DT::datatable()

tabla34


# Cant_calles = 5
# varia_calles34 <- tabla34$x$data[[2]]
# 
# calles34<-varia_calles34[1:Cant_calles]


#tabla34$x$data[[2]][1]

```



Col
-----------------------------------------------------------------------


### **Comparación calles más congestionadas: niveles 3 y 4**





























































```{r}
#varia20 <- varia_calles[1:20,]

Cant_calles <- 5
calles34<-varia_calles34[1:Cant_calles,]

rank34 <- ggplot() +               
  geom_bar(data = varia_calles34, aes(x=reorder(street,sem1), y=sem1),stat = "identity", position = position_dodge(),fill = "dark orange",alpha=0.5)+
  geom_bar(data = varia_calles34, aes(x=street, y=sem2),stat = "identity", position = position_dodge(),alpha=0.4)+
    coord_flip() +
  theme_minimal()+
  labs(x= NULL, y = "Cantidad de eventos")  #+
  #ggtitle("Ranking de calles más congestionadas según semana anterior")

ggplotly(rank34)
```


```{r include=FALSE}

#bb34 = df_aux34 %>% inner_join(aa34 %>% dplyr::select(street))



#tt = datJam %>% inner_join(varia_calles34 %>% dplyr::select(street))


desc_sem2 <- datJam %>%
  inner_join(calles34 %>% dplyr::select(street)) %>%
  filter(sem == 'sem2') %>%
  group_by(street,nombreDia,hora) %>%
  dplyr::summarise(Cant=n())

semt<-datJam %>%
  group_by(sem,nombreDia,hora) %>%
  dplyr::summarise(Cant=n())

d_h34 <- ggplot(desc_sem2) +
  aes(x = hora, y = Cant, fill = street) +
  geom_bar(stat = 'identity',position = 'dodge') +
  #geom_line() +
  theme_minimal()+
  facet_grid(nombreDia~.)+
  labs(title = "Congestion diaria por hora semana actual vs semana anterior (OTRA VERS)", x = "dia_hora", y = 'Cantidad de eventos', colour = "Cantidad eventos")

calle_linea <- ggplot(desc_sem2) +
  aes(x = hora, y = Cant, group=street, color= street) +
  geom_line() +
  scale_color_brewer(palette="Accent")+
  
  theme_minimal()+
  facet_grid(nombreDia~.)+
  #theme(legend.position = "top") +
  labs(x = "Hora", y = 'Cantidad de eventos')


#ggplotly(d_h34)
#ggplotly(calle_linea)

fig <- ggplotly(calle_linea)


# fig <- a %>% layout(legend = list(orientation = 'h')) %>%
#   layout(legend= list(itemsizing='constant'))

#legendOptions(position = c('topleft')) 
```






























```{r}

pp2 <- datJam %>%
  inner_join(calles34 %>% dplyr::select(street)) %>%
  filter(sem == 'sem2') %>%
  group_by(street,level,length,speedkmh,delay)  %>%
  ungroup() %>%
  group_by(street) %>%
  dplyr::summarise(Cant=n(), 
                   Media_d= round(mean(delay),0),
                   q10_d = quantile(delay, prob=c(0.10)),
                   q90_d = quantile(delay, prob=c(0.90)),
                   Media_vel= round(mean(speedkmh),0),
                   q10_vel = round(quantile(speedkmh, prob=c(0.10)),0),
                   q90_vel = round(quantile(speedkmh, prob=c(0.90)),0),
                   Media_len= round(mean(length),0),
                   q10_len = quantile(length, prob=c(0.10)),
                   q90_len = quantile(length, prob=c(0.90)))
  

fig_delay<- ggplot(pp2) +
  aes(x=street, y=Media_d) + 
  #geom_line() +
  geom_point(colour="darkgreen",size = 3)+
  geom_errorbar(aes(ymin=q10_d, ymax=q90_d), width=.2,
                 position=position_dodge(0.05)) +
   coord_flip() +
  theme_minimal()+
  labs(x= NULL, y = NULL)

fig_vel<- ggplot(pp2) +
  aes(x=street, y=Media_vel) + 
  #geom_line() +
  geom_point(colour="darkorange",size = 3)+
  geom_errorbar(aes(ymin=q10_vel, ymax=q90_vel), width=.2,
                 position=position_dodge(0.05)) +
   coord_flip() +
  theme_minimal()+
  labs(x= NULL, y = NULL)

fig_len<- ggplot(pp2) +
  aes(x=street, y=Media_len) + 
  #geom_line() +
  geom_point(colour="darkgrey",size = 3)+
  geom_errorbar(aes(ymin=q10_len, ymax=q90_len), width=.2,
                 position=position_dodge(0.05)) +
   coord_flip() +
  theme_minimal()+
  labs(x= NULL, y = NULL)



pp3 <- datJam %>%
  inner_join(calles34 %>% dplyr::select(street)) %>%
  filter(sem == 'sem2') %>%
  group_by(street,level)  %>%
  dplyr::summarise(Cant=n())


fig_nivel<- ggplot(pp3) +
  aes(x=level, y=Cant, fill=street) + 
  geom_bar(stat = 'identity',position = 'dodge')+
  theme_minimal()+
  #coord_flip() +
  labs(title='Cantidad de eventos por nivel', x = "level", colour = "Cantidad eventos")


#ggplotly(fig_delay)

```


















































































**Calles más congestionadas**
=======================================================================


Column {data-width=700}
-----------------------------------------------------------------------

### **Eventos por día y hora de las 5 calles más congestionadas la semana anterior con niveles 3 y 4**

```{r}
fig
```


Column {data-width=300}
-----------------------------------------------------------------------


### Delay en segundos

```{r}
ggplotly(fig_delay)
```

### Velocidad en Km/h


```{r}
ggplotly(fig_vel)
```

### Largo en metros

```{r}
ggplotly(fig_len)
```